EuroHOPE
Building register-based performance indicators for
ACS and AMI using individual-level administrative health care data
Version of August 27, 2016
BRIDGEHEALTH WP11 Integrating data sources into a comprehensive EU Information System for Health Care Monitoring and Reporting
2
Contents
1. INTRODUCTION AND OBJECTIVES ........................................................................................................................... 3
2. CONSTRUCTION OF DATA ....................................................................................................................................... 4
NATIONAL DATABASES OF AMI AND ACS ................................................................................................................................ 5
NATIONAL COMPARISON DATABASE FOR CALCULATING INDICATORS ............................................................................................... 6
3. HOSPITAL AND FIRST HOSPITAL EPISODE ................................................................................................................ 7
DEFINITION OF A HOSPITAL .................................................................................................................................................... 7
DEFINITION OF THE FIRST HOSPITAL EPISODE ............................................................................................................................. 8
REHABILITATION .................................................................................................................................................................. 8
LENGTH OF STAY, ACUTE CARE AND NON-ACUTE CARE ................................................................................................................. 9
HOSPITAL HIERARCHY ........................................................................................................................................................... 9
4. DESCRIPTION OF INDICATORS ................................................................................................................................11
BASELINE PATIENT CHARACTERISTICS ..................................................................................................................................... 12
PROCESS INDICATORS ......................................................................................................................................................... 13
OUTCOME INDICATORS ....................................................................................................................................................... 14
ADJUSTING FOR PATIENT MIX ............................................................................................................................................... 15
LEVELS OF ANALYSIS ........................................................................................................................................................... 16
REFERENCES: .............................................................................................................................................................17
APPENDIX 1. NATIONAL REGISTERS AND DATA SOURCES USED IN NATIONAL DATABASES .......................................18
APPENDIX 2. PROCEDURE CODES USED IN COUNTRIES TO IDENTIFY PROCEDURES IN TREATMENT OF ACS/AMI ......19
APPENDIX 3. REGIONS USED IN REPORTING OF INDICATORS IN EUROHOPE COUNTRIES ..........................................21
APPENDIX 4. GUIDELINES AND STEPS FOR BUILDING THE NATIONAL ACS/AMI DATABASE .......................................22
3
1. Introduction and objectives
The principal aim of the BRIDGE Health Work Package 11 “Integrating data sources into a
comprehensive EU Information System for Health Health Care Monitoring and Reporting” is to
create databases to enable comparison of performance in the care of specific patient groups
between countries, within countries (regions and hospitals), and over time, using patient-level
administrative health care data. The specific aims are updating the protocols, data processing,
and reporting for selected diseases/conditions included in the European Health Care Outcomes,
Performance and Efficiency (EuroHOPE) project. This paper updates the earlier version of the
protocol for acute myocardial infarction (AMI)1, which has been applied in making of the
database that has been utilized in several articles (Hagen et al. 2015, Hagen et al. 2015b,
Häkkinen et al. 2015) as well as in the making of regional indicators available at
www.eurohope.info. In addition, acute coronary syndrome (ACS) is now added to the protocol
and the protocol has been revised accordingly.
In the earlier stage of EuroHOPE, the AMI data was gathered from Finland, Hungary, Italy,
Netherlands, Scotland and Sweden for the years 2006-2008 and from Norway for 2009. Now the
data will be updated for Finland, Hungary, Italy, Norway, and Sweden to cover more recent
years. In addition, data from Denmark and Spain (Madrid) will be compiled as new entrants to
the comparison.
The main objective of the comparison database is to produce performance indicators at country,
regional and hospital level for international benchmarking. The database enables to extend and
deepen the international comparative research on the relationship between outcomes/quality
and costs/resources as well as on the reasons behind the differences in outcomes and costs
(Hagen et al. 2015, Häkkinen et al. 2015).
This specific protocol for international comparisons for ACS/AMI describes how the EuroHOPE
international comparison data is constructed when based on hospital discharge registers,
mortality registers, and other available administrative health care registers (such as data on
medication use, specialist visits etc.). The protocol is used for preparing both the national ACS
and AMI databases for each country and for international comparative ACS and AMI
databases, which are produced using the national databases (Figure 1).
This protocol also defines how we have produced indicators on ACS and AMI at national level
and also on regional- and hospital-level within countries. The indicators include basic
information on patients (number of patients, demographic characteristics, co-morbidity), on the
1The paper was a joint work established by Eva Belicza, Anne Douglas, Terje P. Hagen, Unto Häkkinen, Amber van der
Heijden, Antti Malmivaara, Emma Medin, Dino Numerato, Mikko Peltola, Clas Rehnberg and Timo T. Seppälä. Éva Belicza was the first author.
4
content of care (use of services and procedures, costs, treatment practices, process indicators)
and outcomes (mortality, recurrence, rehospitalisation, complications).
The protocol has been updated to be applied in the present project. Participants in the present
project are:
- University of Southern Denmark, Odense, Denmark
- National Institute for Health and Welfare, Helsinki, Finland
- Centre for Research on Health and Social Care Management, Università Commerciale Luigi
Bocconi, Milan, Italy
- Health Services Management Training Centre, Semmelweis University, Budapest, Hungary
- Instituto de Salud Carlos III (ISCIII), Madrid, Spain
- Medical Management Centre, LIME, Karolinska Institutet, Stockholm, Sweden
- Department of Health Management and Health Economics, University of Oslo, Oslo,
Norway.
Figure 1. Schematic presentation of data flow in BridgeHEALTH WP11.
2. Construction of data
We will analyse separately patients suffering from acute myocardial infarction (AMI) and acute
coronary syndrome ACS. ACS extends AMI to include patients suffering from unstable angina
pectoris (UAP) because the increased use of diagnostic procedures has affected the number of
Non–ST-segment elevation myocardial infarction (NSTEMI) patients (earlier defined as UAP
patients).
5
ACS and AMI are analysed separately in order to guarantee comparability to other studies on AMI
and earlier reporting of AMI in EuroHOPE.
National databases of AMI and ACS
Total incidence of ACS/AMI in a given calendar year comprises of all patients admitted to hospital
due to ACS/AMI and persons who have died of coronary artery disease without being admitted to
hospital. ACS/AMI may be fatal and the person may not reach a hospital. Partly the access to
treatment may depend on the local health care system characteristics. In EuroHOPE we try to
assess the number of persons who suffered from ACS/AMI irrespective of the access to hospital
care. Persons who died of coronary artery disease without being admitted to hospital due to
ACS/AMI are gathered from countries where available. The analysis of total incidence of AMI/ACS
will be explored later. However, the health system’s performance in treatment of ACS/AMI is
assessed by analysing the persons being treated in hospital due to these diseases.
In EuroHOPE, every country has established national AMI and ACS databases, that includes
patients treated in hospital due to AMI/ACS (prevalence of the diseases in acute care).From
national discharge registers all patients that had been admitted to hospital inpatient care because
of main diagnosis of AMI (International Classification of Diagnoses, 10th edition [ICD-10] codes
I21*, I22*; ICD-9 codes: 410*) or ACS (ICD-10 codes I20.0*, I21*, I22*; ICD-9 codes: 410*, 4111*)
were included in the databases.
AMI cases are classified into three subgroups2:
1. STEMI or recurrent AMI (ICD-10 codes: I21.0*, I21.1*, I21.2*, I21.3*, I22*; ICD- 9 codes:
4100*-4106*, 4108*, 41072, 41092).
2. Undefined AMI (ICD-10 code I21.9*; ICD-9 code: 4109* excluding if fifth digit is 2).
3. NSTEMI (ICD-10 code I21.4*; ICD-9 code: 4107* excluding if fifth digit is 2).
For ACS, a fourth subgroup is introduced:
4. Unstable angina (ICD-10 code: I20.0*; ICD-9-code: 4111*).
Using unique and linkable personal identification numbers, we have linked AMI and ACS patients’
information from the following national registers:
- Hospital discharge registers
- Outpatient services in specialist care / hospitals
- Drug utilisation registers
- National mortality registers.
2 In Hungary, AMI cases can not be reliably classified into these subgroups 1 to 3.
6
National comparison database for calculating indicators
For an explanation regarding the approach used in this part of the study, please see Häkkinen et
al. (2013).
Registry data on hospital discharges, prescription drugs and causes of death were acquired in the
participating European countries. This chapter describes in detail how the 2013 cohort of the
national ACS/AMI comparison data in EuroHOPE was created, starting from the prevalence of
ACS/AMI in acute hospital care. Datasets covering other cohorts are created using the same logic.
The steps in constructing the Finnish national comparison data are also shown in flow charts in
Appendix 4.
First, using hospital discharge data all patients admitted between 1st January 2013 and 31st
December 2013 with a main diagnosis of ACS/AMI were identified. The hospital discharge records
and all the identified patients’ records in the other data sources mentioned above were gathered
for the period between 1.1.2012 and 31.12.2014, i.e. for the preceding and following calendar
years in addition to the cohort year data. The first ACS/AMI admission (index admission) of the
year was identified as it starts the follow-up of the patient.
Patients with an ACS/AMI admission during the previous 365 days before the index admission
were excluded from the 2013 cohort (ACS/AMI admission = hospital discharge record with an
ACS/AMI diagnosis as the main diagnosis).
For each patient all continuous hospital treatment (the first hospital episode) starting from the
first ACS/AMI admission (index admission) in 2013 was constructed by combining all consecutive
hospital stays for each patient. The consecutive hospital stays need not be in the same hospital,
i.e. hospital transfers are taken into account when making the first hospital episode.
In case a patient had different ACS/AMI subtypes during the first hospital episode, the most
‘severe’ diagnosis was chosen to characterize the condition of the patient. For this purpose, the
following hierarchy of ACS/AMI subtypes was applied (from the most to the least severe)3: STEMI
or recurrent AMI, undefined AMI, NSTEMI, unstable angina. The most severe diagnosis was chosen
as the ACS/AMI subtype characterizing the first hospital episode.
The included patients were followed for up to 365 days from the first day (index day) of the index
admission for inpatient and outpatient care in hospitals, medication purchases and vital status. In
addition, the hospital discharges and use of prescribed medicines in the 365 days prior to the start
of the index admission were used in assessing the presence of comorbid diseases among the
patients.
3 In Hungary it was not possible to identify different AMI subtypes.
7
In each country, patients under 18 years of age, tourists, visitors and other residents with
incomplete personal identification numbers as well as patients with incomplete data on look-back
and/or follow-up period of 365 days were excluded from the national comparison data4.
The main analysis will be done using the patient data collected from the national discharge
registers as described above5. Specific information on registers in each country is provided in
Appendix 1 and on country specific procedure codes in Appendix 2. Appendix 3 gives a
characterization of the classification of regions used in the project. Variable definitions, together
with definitions of comorbid conditions, procedures, complications and hospital hierarchy are
described in a separate excel file.
3. Hospital and first hospital episode
Definition of a hospital
A hospital is a health care institution providing treatment for a number of medical conditions by
specialized staff and equipment. In the present project, we speak of hospitals meaning institutions
providing somatic (non-psychiatric) inpatient care for patients staying overnight (for at least one
night, i.e. inpatients), and usually also health care services (diagnosis, treatment, or therapy) for
patients without staying overnight (i.e. outpatients). A hospital may be a single building or a
number of buildings on a campus. Also, in some countries a hospital can consist on many buildings
in a certain geographical area. For example, in Finland after reorganization of Helsinki University
Hospital in 2006, it includes several buildings in different municipalities in the capital area.
At hospital level analysis we have specified the definition of a hospital in order to be sure that we
are comparing units with a similar structure and scope. For this end, we have formulated a
definition of hospital, and a corresponding classification of different types of hospitals. We have
used these definitions of hospitals in a specific variable depicting the type of care that the patient
has received for each day of the follow-up daily information (during one year follow-up). In
addition, we will gather more detailed information on the hospitals that have the main
responsibility for the care. The more specific hospital-level information collection is to be gathered
for the hospitals acting as the first hospitals in the care chain, and for the hospitals taking the
responsibility of the patient in the first hospital episode (in the individual level data the hospitals
are given variables named FSTHOSP and HEPHOSP, respectively). Thus, FSTHOSP is the hospital
where the patient was initially admitted in. HEPHOSP is defined as the hospital that was highest in
the hierarchy of hospitals which treated the patient during the first week6.
4 In Hungary, patients being imprisoned are excluded as their use of health care services is not included in the hospital
discharge register. 5 In Finland we have excluded patients that had been only in health centers and other hospitals in departments
without specialty code or specialty code is general medicine (Appendix 4). 6 According to data of five countries (Häkkinen et al. 2015) about 13 % of the patients are transferred to a higher-level
hospital within the first week of hospitalization due to AMI.
8
Definition of the first hospital episode
The total episode of care is defined as the entire treatment pathway from the beginning of the
disease to the end of the treatment throughout any hospital admissions, other health service
provisions or purchased medication in order to solve the health problem at hand in a specified
time frame (Figure 2).
The first hospital episode covers all care given to patients as an inpatient in a hospital.
Consecutive hospital discharges are included in the same hospital episode if the preceding hospital
stay’s discharge date is the same as the following discharge’s admission date or the admission
date is the next date after the preceding discharge date. If the patient is immediately transferred
to a rehabilitation centre at the hospital this is included in the first hospital episode (Häkkinen et
al. 2013). The first hospital episode ends when the patient is discharged to home (and is at home
for at least one day), to a nursing home or to a long-term care institution, or the patient dies. The
total episode of care was defined as the entire treatment pathway from the beginning of the
disease (i.e. acute stage of the disease) to the end of the episode (predefined follow-up time, see
below), irrespective of any organizational boundaries (Figure 2).
Figure 2. A schematic presentation of the follow-up of patients throughout the treatment pathway demonstrating the definitions of the first hospital episode and the total episode of care.
Rehabilitation
In some countries (e.g. in Finland) it is difficult to separate rehabilitation given in a hospital from
acute care as well as to separate rehabilitation from long-term care. Some countries (e.g. Hungary)
Admission to ward A
Procedure/treatment in ward A
Admission to ward B
Discharge to another hospital
Outpatient visit
Medication purchase
Total episode of care
First hospital episode
time
Discharge home
9
may have data on all inpatient rehabilitation. Other countries usually have data on inpatient
rehabilitation given in hospitals but no data on rehabilitation given in a specialized rehabilitation
centre.
We have divided the first hospital episode to acute and non-acute care. In countries where
rehabilitation is included in hospital inpatient data and can be separated from acute care this is
coded in a STATE variable7. In addition, an own class in the hospital hierarchy is given for geriatric
wards of hospitals.
We will include inpatient rehabilitation and thus keep our definition of the end of an episode. In
addition, in countries where rehabilitation is included in hospital inpatient data and can be
separated from acute care this will be coded like mentioned earlier. In addition, an own class in
the hospital hierarchy will be given for geriatric wards in hospitals.
Length of stay, acute care and non-acute care
We measured the length of stay (LoS) in acute care during the first hospital episode from the index
day at the start of the initial admission to the last day of acute hospital care during the period of
continuous acute hospital treatment (LoS = last date in acute treatment – index date +1).
We defined acute hospital care as treatment given in a hospital’s intensive care unit, or in other
acute care settings (all medical and surgical specialties). In addition, we calculated several other
LoS measures including the length of the first admission, the total length of the continuous
episode of care, the number of days in rehabilitation during the first continuous episode of care,
and the number of days in hospital during the entire follow-up year. All LoS measures were
truncated at 365 days if the length of stay was longer.
Hospital hierarchy
The daily STATE variable describes in which place or state the patient is. It is based on the idea that
a patient can only be in one place in each day and that with hospital discharge data the days in
institutions can be located in time. In case of overlapping admissions, the STATE variable is marked
with the hospital being in the highest step of hospital hierarchy (defined by each country). In
descending order, the hospitals, institutions or units in the hierarchy are university hospitals,
specialist hospitals, central or regional hospitals and general or local hospitals, rehabilitation,
geriatric and general care, psychiatric care, and long term care.
1. University hospital
7 The daily STATE variable conveys information on if a patient was, in a given day, in a hospital or not, about the type
of hospital where the patient was that date, whether the main diagnosis was related to a certain disease, information about the intensity of the treatment (i.e. acute care, non-acute etc. based on information known about the ward giving the treatment). Thus, the state variables give a possibility to extract and pinpoint the days the patient spent in rehabilitation, even within the first hospital episode or any other hospital stay during the follow-up. The codes for state variables are given in a separate excel-file.
10
A university hospital (teaching hospital) combines hospital treatment to patients with teaching to
medical students and nurses and usually it is linked to a medical school, or university. University
hospital has an extensive array of specialties and services, and university hospitals are able to
provide treatment to the most demanding medical conditions and are responsible for the
treatment of rare and severe medical conditions in their region. University hospitals are usually
tertiary referral hospitals: Tertiary care is specialized consultative health care, usually for
inpatients and on referral from a primary or secondary health professional, in a facility that has
personnel and facilities for advanced medical investigation and treatment, such as a tertiary
referral hospital (Healy, Mckee 2002). Examples of tertiary care services are cancer management,
neurosurgery, cardiac surgery, plastic surgery, treatment for severe burns, advanced neonatology
services, palliative, and other complex medical and surgical interventions.
2. Specialized hospital
Types of specialized hospitals treat certain disease categories such as cardiac, oncology, or
orthopedic problems, and so forth. A specialized hospital may have smaller volumes, but they are
considered to have an excellent know-how in their field.
3. Central or regional hospital
A central hospital is typically the major health care facility in its region, with a fairly large numbers
of beds for intensive care and many specialized facilities (for example surgery, plastic surgery,
childbirth, bioassay laboratories, and so forth).
4. General/local hospital
General hospital is set up to deal with many kinds of disease and injury, and it has an emergency
department to deal with immediate and urgent threats to health. These hospitals have usually
only the basic specialties such as surgery, internal medicine, deliveries and gynecology, ear, nose
and throat disease etc.
5. Rehabilitation
Here we include all rehabilitation given in special rehabilitation hospitals/clinics as well as all other
hospitals if this can be separated from the acute care using diagnoses, procedures, DRGs, or the
department level information. Thus if rehabilitation is given e.g. in a university hospital and it can
be separated from the acute care, the state variable is coded to give information about this.
6. Geriatric and general care
Care given in geriatric wards and care given in general medicine departments, independent of the
hospital type (any of the above accepted care).
7. Psychiatric care
Care given in psychiatric specialties, or having ICD-10 code F* as main diagnosis.
11
8. Long term care
All inpatient care given in nursing homes and other long-term institutions.
4. Description of indicators
The EuroHOPE project aims at constructing a number of indicators describing the performance of
the health care system in treatment of AMI/ACS. With the national comparison data a number of
national-, regional- and hospital-level indicators are produced. The calculation of indicators and
the reporting of the data are based on a common script, executed in Stata on the national
comparison data. Below in Table 1, the indicators that are published on the national- and regional-
level in the EuroHOPE website in the ATLAS tool are described. Here we give indicators only for
AMI patients. The name of the indicator, a short description of the indicator, and the factors used
in risk-adjustment are given in Table 1.
Table 1. EuroHOPE indicators on AMI publicly available on www.eurohope.info.
Indicator Description Risk-adjustment
Number of patients Number of patients included in
the national comparison data.
Number of patients per
100 000 inhabitants
Number of patients included in
the national comparison data
per 100 000 inhabitants.
Age Average and median age of the
patients. Age in years at the
start of the hospital care for
stroke.
Males Share of males.
PCI rate within two days The share of AMI patients
received percutaneous
coronary intervention (PCI)
during the first two days
PCI or CABG rate within 30 The share of patients received
either percutaneous coronary
Age * , sex
12
days intervention (PCI) or coronary
artery bypass surgery (CABG)
during the first 30 days
Length of stay, first hospital
episode
The number of days in acute
hospital treatment during the
first hospital episode.
Consecutive hospital stays are
taken into account when
constructing the first hospital
episode.
Age*, sex
Length of stay, first year The number of days in hospital
treatment during 365 days
after the start of the acute
hospital treatment due to AMI.
Age*, sex
7-, 30-, 90-day and 1-year
mortality
The share of AMI patients who
died within the given period of
time after the start of the first
hospital admission because of
ischaemic stroke.
Age*, sex
Readmission in 30 days
Readmission to acute hospital
care within 30 days after the
end of acute care in the first
hospital episode.
Age*, sex
* Classified: 18-39, 40-44, 45-49, 50-54, 55-59, 60-64, 65-69, 70-74, 75-79, 80-84, 85-89, 90-.
Baseline patient characteristics
In addition to the publicly reported indicators given in Table 1, a number of other indicators are
produced in EuroHOPE. The indicators can be classified as indicators related to the baseline
patient characteristics, process, and outcome.
As baseline patient characteristics the following information is gathered:
- Age and gender
- Type of ACS/AMI
- Comorbidities (see Excel file on variable definitions)
o Hypertension o Coronary artery disease o Atrial fibrillation
13
o Cardiac insufficiency (heart failure) o Diabetes mellitus o Atherosclerosis o Cancer o COPD and asthma o Dementia o Depression o Parkinson's disease o Mental disorders o Renal insufficiency (failure) o Alcoholism o Stroke
Co-morbidities are defined from various register sources according to two different approaches:
1. Based on the main and secondary diagnoses of all hospital inpatient and outpatient
records during the 365 days preceding the index admission
2. Based on medicine purchases and the main and secondary diagnoses of all hospital
inpatient and outpatient records during the 365 days preceding the index admission.
Process indicators
The patients’ first hospital episode and the whole follow-up of one year are tracked for a number
of aspects that convey information about the care given to the patient. The process indicators
produced in the project are the following:
- Length of stay (LoS) of first hospital admission, days per patient
- Length of stay of the first hospital episode, days per patient, in four categories
o Total LoS o Days in acute care o Days in non-acute care o Days due to ACS/AMI (days with main diagnosis of ACS/AMI)
- The number of inpatient days per patient over the first year after ACS/AMI
o Total LoS o Days in acute care o Days in non-acute care o Days due to ACS/AMI (days with main diagnosis of ACS/AMI)
- Number and share of patients who received percutaneous coronary intervention (PCI)
during the first two days
- Number and share of patients who received coronary artery bypass surgery (CABG)
during the first two days
- Number and share of patients who received percutaneous coronary intervention (PCI)
during the first hospital episode
14
- Number and share of patients who received coronary artery bypass surgery (CABG)
during the first hospital episode
- Number and share of patients who received percutaneous coronary intervention (PCI)
during the first 30 days
- Number and share of patients who received coronary artery bypass surgery (CABG)
during the first 30 days
- Number and share of patients who received either percutaneous coronary intervention
(PCI) or coronary artery bypass surgery (CABG) during the first 30 days
- Number and share of patients who used drugs (outside hospitals) based on ATC
classification (anatomic therapeutic classification) one year before and one year after
hospitalisation (these will be grouped later):
o Beta blockers (C07*) o Diuretics (C03*, C07BB*, C09BA*, C09DA*) o ACE inhibitors (C09A*, C09B*) o Angiotensin receptor blockers (AT II antagonists) (C09C*, C09D*) o Calcium channel blockers (C08*, C07FB*, C09BB*) o Insulin (A10A*) o Blood glucose lowering drugs, excluding insulins (oral diabetes medication) (A10B*) o Statins (C10AA*) o ADP receptor inhibitors (clopidogrel, prasugrel, ticagralor) (B01AC04, B01AC22,
B01AC24) o Dipyridamole (B01AC07, B01AC30) o Warfarin (B01AA03) o Antidepressants (N06A*) o Anti-dementia drugs (N06D*) o Antiepileptics (N03A*) o Acenokumarol (B01AA07) o Ticlopidin (B01AC05) o Dabigatran (B01AE07) o Apixaban (B01AF02) o Rivaroxaban (B01AF01)
Outcome indicators
The project aims at constructing measures to be used for performance monitoring and assessing
the outcomes of care given to the patients. As outcome indicators, the following measures are
included:
- Mortality in 30, 90, and 365 days from the index admission day
- Readmission (due to recurrence of ACS or AMI) to hospital within
o 30 days o 90 days o 365 days from the index admission
- Readmission to acute hospital care within 30 days after end of the first hospital episode
15
- Readmission to acute hospital care within 30 days after end of the acute hospital care in
the first hospital episode
- Complications during the first hospital episode:
o pulmonary embolism o stroke
Adjusting for patient mix
Comparisons of health outcomes between countries need to take into account differences in the
patient mix. In addition, countries may differ in the degree to which the relevant information is
recorded, the availability of patient information, or variables being very differently defined across
countries. In order to the performance indicators to be comparable, the indicators have to be
adjusted for confounding factors.
In EuroHOPE this problem was tried to solve by using all relevant registry data available for
everyone with a specified health problem, by collecting available information on disease specific
comorbidities, length of hospital stay and medication use prior to the occurrence of the health
problem studied - variables potentially having an effect on health outcomes. However, this does
not alleviate the problem arising from the potential existence of differences between countries in
registering this information.
Three different risk-adjusted outputs are produced for each outcome:
1. adjusted for sex and age, and unstable angina pectoris (UAP)8
2. adjusted for sex, age, disease-specific comorbidities based on primary and secondary
diagnoses9, the number of hospital days (LOS) the year prior to index admission and
unstable angina pectoris (UAP)8
3. adjusted for sex, age, disease-specific comorbidities based on primary and secondary
diagnoses and medication purchases, LOS the year prior to index admission and unstable
angina pectoris (UAP)8.
Based on the experiences in the PERFECT project (Peltola et al., 2011), the observed/expected
approach described by Ash et al. (2003) is used - this roughly corresponds to indirect
standardization. Specifically, the method uses regression modelling for the risk adjustment. For
mortality outcomes up to one year, logistic regression is used, while for the LOS outcomes,
negative binomial regression is used. In each country, a common indicator-specific set of
coefficients for each factor included in the risk-adjustment is used for calculation the predicted
values for the outcome in question. The coefficients applied for calculating the predicted values
for each outcome are based on the estimates acquired from the Finnish national comparison data
8 UAP is only used in risk adjustment for ACS.
9 Hypertension, coronary heart disease, atrial fibrillation, cardiac insufficiency, diabetes mellitus, cancer, chronic
obstructive pulmonary disease and asthma, dementia, depression, Parkinson's disease.
16
covering the years 2006 to 2013. The coefficients will be updated as data from other countries is
available. The method is described in greater detail in Moger and Peltola (2014).
Each country will apply a standardized, centrally-constructed, Stata syntax code to the national
comparison database for calculating the country and regional level indicators. The national files
were processed with a common script in order to enable standardized reporting of the data from
all countries with minimum workload and minimized possibility of human error in processing the
data. This Stata do-file is available upon request from the researchers.
Case-mix standardisation will be used when comparing countries, regions, hospitals, or years.
Variables which are considered potential prognostic factors (and thus confounders) are used for
adjustment. These will be derived from primary and secondary diagnoses of previous discharge
data and from data on previously prescribed medicines. We will use the following variables:
- age (in years, classified)
- gender
- type of ACS/AMI: unstable angina pectoris (UAP)8
- comorbidity as defined in separate file (only the comorbid diseases with at least 1%
prevalence in the study population in each country of the EuroHOPE partners’ data in the
year 2007 were included in the risk adjustment as confounding factors)
- inpatient hospital stay days during one year prior to ACS/AMI in acute inpatient hospital
care.
Levels of analysis
Indicators are produced at national and also regional level within countries and later at hospital
level. Regional information is based on patients’ place of residence. The definitions for regions
have been made in each country according to the local preferences (Appendix 3).
17
References:
Ash AS, Schwartz M, Peköz EA. Comparing outcomes across providers. In Risk Adjustment for
measuring health care outcomes. 3rd ed. Edited by Iezzoni LI. Chicago: Health Administration
Press; 2003:297-333.
Healy J, McKee M. The role and function of hospitals. In Hospitals in a changing Europe: McKee M,
Healy J. (eds.); 59-80. Open University Press: Buckingham. 2002.
Häkkinen U, Iversen T, Peltola M, Seppälä TT, Malmivaara A, Belicza E, Fattore G, Numerato D,
Heijink R, Medin E, Rehnberg C. Health care performance comparison using a disease-based
approach: The EuroHOPE project. Health Policy 2013;112(1-2):100-109.
Häkkinen U, Rosenqvist G, Seppälä TT, Iversen T, Rehnberg C, and on behalf of the EuroHOPE
study group. Outcome, use of resources and their relationship in the treatment of AMI, stroke
and hip fracture at European hospitals. Health Economics, 2015; 24 (Suppl. 2): 116-139.
Malmivaara A, Meretoja A, Peltola M, Numerato D, Heijink R, Engelfriet P, Wild SH, Belicza É,
Bereczki D, Medin E, Goude F, Boncoraglio G, Tatlisumak T, Seppälä TT and Häkkinen U.
Comparing ischaemic stroke in six European countries. The EuroHOPE register study. Eur J
Neurol. 2015; 22(2): 284-91, e25-6.
Moger TA, Peltola M. Risk adjustment of outcomes in a multinational register-based study – A
pragmatic approach to a complicated topic. SAGE Open Medicine January - December 2014
vol. 2.
Peltola M, Juntunen M, Häkkinen U, Rosenqvist G, Seppälä TT, Sund R. A methodological approach
for register-based evaluation of cost and outcomes in health care. Annals of Medicine 2011;
43: S4-S13.
Peltola M, Seppälä TT, Malmivaara A, Belicza É, Numerato D, Goude F, Flethcher E, Heijink R, and
on behalf of the EuroHOPE study group. Individual and Regional Level Factors Contributing to
Variation in Length of Stay after Cerebral Infarction in Six European Countries. Health
Economics, 2015; 24 (Suppl. 2): 38-52.
18
Appendix 1. National registers and data sources used in national databases
Hospital discharge register for inpatient care
Denmark 2005-2014
Finland 2005-2014
Italy
Hungary 2005-2015
Norway 2008-2015
Spain
Sweden 2005-2014
Register on use of outpatient services in hospitals and/or other specialist units
Denmark 2005-2014
Finland 2005-2014
Italy
Hungary 2005-2015
Norway
Spain
Sweden
Register on prescribed medication
Denmark 2005-2014
Finland 2005-2014
Italy
Hungary 2005-2015
Norway 2004-2015
Spain
Sweden 2005-2014
Causes of death Denmark 2005-2013
Finland 2005-2014
Italy
Hungary NA (dates of death available for 2005-2015)
Norway 2004-2015
Spain
Sweden 2005-2014
19
Appendix 2. Procedure codes used in countries to identify procedures in treatment of ACS/AMI
ACS and AMI Codes
OPE Procedure Denmark Finland Hungary Italy Norway Spain Sweden
ANG Angiography (UXAC85) FN1AC, FN1BC, FN1CC, XFN00, 61*, 81*, AN1**, AP1**
33300, 33304, 33305, 33306, 33307, 33308, 33340, 33341, 33345, 33346, 33347, 33348, 33351, 33353, 33360, 33361, 33362, 33400
TFC00, TFC10, XF911, XF913, FYDB12, FYDB13
AF001, AF002, AF003, AF004, AF005, AF006, AF037
CABG Coronary artery bypass surgery
KFNA*, KFNC*, KFNE*, KFNF*
FNA, FNB, FNC, FND, FNE, 11***, 25***, 111****, 112****, 113****, 119****, AA1**, AA2**, AA3**, AAX**
53611, 53612, 53613, 53614, 53615, 53616, 53617
FNA, FNB, FNC, FND, FNE
PDH10, PDH30, PEH10, PEH11, PEH12, PEH20, PEH30, PCH10, PCH20, PCH30, PCH40, PCH99, PBH10, PBH20, PBH99, PAH10, PAH20, PAH21, PAH25, PAH30, PAH99, FAD00, FAD10, FAD96, FCC70, FCC76, FCD70, FDH40, FJD20, FNC10, FNC20, FNC30, FNC40, FNC50, FNC60, FNC96, FND10, FND20, FND96, FNE00, FNE10, FNE20,
20
FNE96, FNH20, PFH10, PFH20, PFH21, PFH22, PFH23, PFH24, PFH25, PFH26, PFH27, PFH28, PFH29, PFH99
PCI Percutaneous coronary intervention
KFNG02, KFNG05, KFNG05A - only acute
FN1AT, FN1BT, FN1YT, TFN40, TFN50, 82*, 83*, 84*, AN2**, AN3**, AN4**, ANA**
01100, 33970, 33971, 33972, 53963, 53966
FNF00, FNF10, FNF20, FNF30, FNF96, FNG00, FNG02, FNG10, FNG20, FNG22, FNG30, FNG96, FNG05
FNG02
PCIST PCI + stent KFNG05, KFNG05A
TN1YT, TFN50, AN3**, AN4**
01344, 33974, 33975, 33976, 33981, 33982, 33983, 33984, 33985, 33986, 33987, 33988, 33989, 3398A
FNG05
PCISTm
PCI + metal stent
As above AN3** 01339
PCISTd PCI + drug eluting stent
As above AN4**
*HDR/Data file on coronary patients -2005 **HDR/Data file on coronary patients 2006- ***HDR/Data file on coronary patients -2002 ****HDR/Data file on coronary patients 2003-2005
21
Appendix 3. Regions used in reporting of indicators in EuroHOPE countries
Country Description Number of regions
Average population size
Finland Hospital districts and hospital regions responsible for providing specialised health care. Smallest districts combined.
19 280 000
Denmark Administrative regions. 5 1 000 000
Hungary 19 counties and Budapest area providing self-governmental administrative duties (not health care).
20 500 000
Italy Counties of the Friulia-Venezia Giulia autonomous region. Counties responsible for providing health care.
4 300 000
Norway Hospital trusts responsible for providing specialist health care in their geographical areas.
20 250 000
Spain
Sweden Counties responsible for providing health care.
21 450 000
22
Appendix 4. Guidelines and steps for building the National ACS/AMI database
This example shows how the Finnish database is formed. However, there will certainly be
differences in each country and thus these steps have to be modified accordingly.
1st step: screening inpatient database for patients
Screen hospital database (hospital discharges/hospital department discharges, inpatient social
care), from the year 2004 onwards for records with ACS/AMI as main diagnosis in all hospital stays
(hospital departments).
2nd step: screening mortality register database for ACS patient treated at hospitals
Take patient IDs from the first step and gather their information on date of death and causes of
death and place of death.
3st step: screening mortality register database for ACS patients not having hospital care (not
possible in all countries)
Screen national mortality database from the year 2006 onwards for records with main diagnosis
(ICD-10: I20-I25, I46, R96, R98, R99; ICD-9: 410, 411, 413, 414, 427, 798.1, 798, 799). Take patient
ID and main diagnosis of death, date of death and place of death.
4th step: merge data
Merge data from steps 1, 2 and 3 together with patient id in order to create an ACS ID data that
includes four elements:
i. patient ID
ii. main diagnosis of death (if available)
iii. place of death (in hospital / outside hospital)
iv. date of death
v. other reasons of patient drop out (eg. moving from the country).
5th step: 1st data set, ACS/AMI (prevalence) (1)
Take patient IDs from the fourth step and gather their all records from hospital records.
6th step: 2nd data set, all patients in hospital care due to ACS/AMI (2)
Exclude all patients that have not been in hospital care in the year under consideration due to
ACS/AMI.
7th step : National comparison data (3)
Make the exclusions given in section 2.
23
Constructing the Finnish databases
Figure A1 describes the construction of Finnish ACS data from the year 2013. Total prevalence of
ACS was 21 129. Of these 9514 (45 %) had not been in hospital because of ACS in 2013.
The Finnish Care Register for Health Care (FCRC) includes data from various hospitals (e.g.
rehabilitation hospitals, health centers). In order to make the patients more comparable with
hospital register in other countries we have excluded patients that had been only in health centers
and other hospitals in departments without specialty code or whose specialty code is general
medicine. This further decreased the number of patients by 737.
Table A1 describes development (2006-2013) of the structure of ACS data. The structure has been
rather stable during time period. The main change has been in the type of AMI. The share of
NSTEMI patients has increased from 32 % to 45 % whereas the share of unstable angina has
decreased from 22 % to 16 %, mainly because extended use of diagnostic procedures has
increased the share of NSTEMI patients (earlier defined as UAP patients). The total share of
NSTEMI and UAP patients has increased from 54 % to 61 %. Extended use of diagnostic procedures
may have also affected to the decrease of undefined AMI patients.
Figure A2 and Table A2 describe similar information on AMI patients in Finland.
24
1st
data set (prevalence). All persons in
institutional care due to ACS or died due to
coronary artery disease 1)
in 2013 n = 21 129 (1)
2nd
data set All patients in acute hospital care due to ACS in 2013
n = 10 878 (2)
National comparison data. Number of ACS patients in the comparison
database after exclusion n = 10 094 (3)
Number of persons who died because of coronary artery
disease and had no hospital care2)
due to ACS in 2013
n = 9 514 (1a)
Number of persons who had no hospital
care or long term care 3)
during the previous 365 days
n = 2 970 (1a1)
Exclusion 1: Hospital admission due to ACS during the
previous 365 days 5)
n = 604 (2a)
Exclusion 2: Persons under 18 years at the time of first
index admission in 2013 n = 0 (2b)
Exclusion 3:
Foreigners 6)
and patients with incomplete ID
n = 180 (38=incomplete ID) (2c)
Number of persons who were in long term care or had hospital care but no acute hospital care
due to ACS and did not die n = 737 (1b)
Number of persons who had acute hospital care due to ACS during the
previous 365 days n = 404 (1a2)
Number of persons who had non-acute
hospital care 4)
or long term care due to ACS during the previous 365 days
n = 191 (1a3)
Number of persons who had hospital or long term care only due to other diagnosis than
ACS during the previous 365 days n = 5 949 (1a4)
Acute hospital care
n = 4 549
(1a4a)
No acute hospital care
n = 1 400 (1a4b)
Stemi n = 2 893
(3a)
Recurrent AMI
n = 43 (3b)
Undefined AMI
n = 991 (3c)
Non-stemi n = 4 544
(3d)
Unstable angina
n = 1 623 (3e)
25
Figure A1. Creation of the Finnish comparison database for ACS in 2013
1. Coronary artery disease is defined according to the following causes of death codes: 'I20' ,
'I21' , 'I22' , 'I23' , 'I24' , 'I25' , 'I46' , 'R96' , 'R98' , 'R99' .
2. Hospital care is defined as inpatient hospital care only. Hospital data is based on the
Finnish Care Register for Health Care (FCRC)
3. Data on long-term care is based on the Care Register for Social Welfare
4. Non acute hospital care includes care given in health centers and other hospitals (included
in FCRC) in departments without specialty code or specialty code is general medicine.
5. Counting starts from the first index admission in 2013.
6. In Finland all the ACS patients whose home municipality is Åland or unknown are excluded
from the comparison database
Table A1. Construction of ACS data bases in Finland 2006-2013
Data 2006 2007 2008 2009 2010 2011* 2012 2013
1 23 304 23 150 22 642 22 255 22 118 18 978 21 570 21 129
1a 10 032 10 197 10 306 10 142 10 424 7 305 9 883 9 514
1a1 3 240 3 283 3 381 3 167 3 296 1 554 3 060 2 970
1a2 597 554 542 440 514 439 416 404
1a3 212 206 220 218 189 189 182 191
1a4 5 983 6 154 6 163 6 317 6 425 5 123 6 225 5 949
1a4a 4 769 4 909 4 854 4 871 4 877 3 976 4 744 4 549
1a4b 1 214 1 245 1 309 1 446 1 548 1 147 1 481 1 400
1b 719 741 834 844 774 739 711 737
2 12 553 12 212 11 502 11 269 10 920 10 934 10 976 10 878
2a 877 768 732 740 691 612 648 604
2b 2 1 1 3 0 0 0 0
2c 104 144 148 164 120 132 132 180
2c1 16 10 4 2 1 1 1 38
3 11 570 11 299 10 621 10 362 10 109 10 190 10 196 10 094
3a 2 891 2 940 3 061 2 803 2 750 2 890 2 840 2 893
3b 308 237 189 166 114 68 63 43
3c 2 101 2 047 1 917 1 939 1 632 1 351 1 238 991
3d 3 729 3 583 3 278 3 309 3 614 3 925 4 263 4 544
3e 2 541 2 492 2 176 2 145 1 999 1 956 1 792 1 623
* Some deaths are missing
26
1st
data set (prevalence). All persons in
institutional care due to ACS or died due to
coronary artery disease 1)
in 2013 n = 21 129 (1)
2nd
data set All patients in acute hospital care due to ACS in 2013
n = 10 878 (2)
National comparison data. Number of ACS patients in the comparison
database after exclusion n = 10 094 (3)
Number of persons who died because of coronary artery
disease and had no hospital care2)
due to ACS in 2013
n = 9 514 (1a)
Number of persons who had no hospital
care or long term care 3)
during the previous 365 days
n = 2 970 (1a1)
Exclusion 1: Hospital admission due to ACS during the
previous 365 days 5)
n = 604 (2a)
Exclusion 2: Persons under 18 years at the time of first
index admission in 2013 n = 0 (2b)
Exclusion 3:
Foreigners 6)
and patients with incomplete ID
n = 180 (38=incomplete ID) (2c)
Number of persons who were in long term care or had hospital care but no acute hospital care
due to ACS and did not die n = 737 (1b)
Number of persons who had acute hospital care due to ACS during the
previous 365 days n = 404 (1a2)
Number of persons who had non-acute
hospital care 4)
or long term care due to ACS during the previous 365 days
n = 191 (1a3)
Number of persons who had hospital or long term care only due to other diagnosis than
ACS during the previous 365 days n = 5 949 (1a4)
Acute hospital care
n = 4 549
(1a4a)
No acute hospital care
n = 1 400 (1a4b)
Stemi n = 2 893
(3a)
Recurrent AMI
n = 43 (3b)
Undefined AMI
n = 991 (3c)
Non-stemi n = 4 544
(3d)
Unstable angina
n = 1 623 (3e)
27
Figure A2. Creation of the Finnish comparison database for AMI in 2013
1. Coronary artery disease is defined according to the following causes of death codes: 'I20' ,
'I21' , 'I22' , 'I23' , 'I24' , 'I25' , 'I46' , 'R96' , 'R98' , 'R99' .
2. Hospital care is defined as inpatient hospital care only. Hospital data is based on the
Finnish Care Register for Health Care (FCRC)
3. Data on long-term care is based on the Care Register for Social Welfare
4. Non acute hospital care includes care given in health centers and other hospitals (included
in FCRC) in departments without specialty code or specialty code is general medicine.
5. Counting starts from the first index admission in 2013.
6. In Finland all the AMI patients whose home municipality is Åland or unknown are excluded
from the comparison database
Table A2. Construction of AMI data bases in Finland 2006-2013
Data 2006 2007 2008 2009 2010 2011* 2012 2013
1 20 765 20 598 20 312 19 963 20 004 16 932 19 680 19 428
1a 10 122 10 294 10 398 10 226 10 509 7 383 9 946 9 584
1a1 3 242 3 289 3 384 3 168 3 298 1 557 3 061 2 971
1a2 563 524 516 412 490 407 400 388
1a3 207 197 215 210 183 183 178 184
1a4 6 110 6 284 6 283 6 436 6 538 5 236 6 307 6 041
1a4a 4 890 5 029 4 968 4 981 4 984 4 081 4 822 4 634
1a4b 1 220 1 255 1 315 1 455 1 554 1 155 1 485 1 407
1b 580 571 626 662 596 597 574 625
2 10 063 9 733 9 288 9 075 8 899 8 952 9 160 9 219
2a 609 557 494 507 484 410 455 432
2b 2 1 1 2 0 0 0 0
2c 86 116 119 131 103 112 118 159
2c1 13 10 3 2 1 0 1 35
3 9 366 9 059 8 674 8 435 8 312 8 430 8 587 8 628
3a 2 959 2 995 3 116 2 855 2 797 2 926 2 875 2 930
3b 321 248 198 171 117 69 64 44
3c 2 157 2 109 1 961 1 980 1 667 1 375 1 257 1 006
3d 3 929 3 707 3 399 3 429 3 731 4 060 4 391 4 648
* Some deaths are missing